Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “statistical change detection”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

RTN-056: Study of the Photon Transfer Curve in the CCD detectors of the Vera C. Rubin Observatory

The RECA internship program provides Colombian students with an opportunity to enhance their research skills in Astronomy, Astrophysics, and Cosmology. During this three-month program, our main objective was to study the Photon Transfer Curves (PTC) of the Vera C. Rubin Observatory, specifically the gain, and to compare it with the gain obtained through pairs of flats. Overall, the study of PTCs is crucial in understanding the performance of detectors and instruments used in Astronomy. The Vera C. Rubin Observatory is an important facility that will enable researchers to carry out a wide range of studies in this field, making it essential to investigate its gain performance. We used run 13144 to construct the PTCs and 13186 to analyze the crosstalk. We employed the LSST Science Pipelines (also known as the DM stack), a software under development for this observatory, which performs all the necessary reductions for the construction of the PTCs. We also used simulations to replicate the observed effects. Initially, we found a 5% difference between the gain calculated by PTC and pairs of flats for a flow range between 5000 and 10000 ADU. Simulations showed that this difference was due to the handling of statistics and the assumption that the distribution following the Lupton equation is Gaussian. We found an error interval for this flow region based on the vendor, with (1.8 ± 0.7, 4.1 ± 0.9) % for E2V and (0.85 ± 0.7, 2.2 ± 0.9) % for ITL. From the PTC, we also obtained the average Full Well Capacity of LSSTCam as 130000 ± 10000$ electrons. We identified a list of segments where we found differences with the results obtained by SLAC National Acceleration Laboratory in PTC parameters, low saturation level, or other defects. We detected and corrected the effect of statistics in the gain calculation using pairs of flats and proposed a code change, which was implemented in the pipeline software. We do not recommend correcting for crosstalk as it does not significantly affect the parameters and does not change the shape of the PTC. However, the opposite is true for the nonlinearity correction.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2020

Data-driven occupant-behavior analytics for residential buildings

Many advances have been made in building technology to help save energy, but influencing the behavior of the occupants is still necessary to achieve low-energy use targets. One of the most practical ways to influence and change occupant behaviors is through incentives. Developing incentives for energy-saving and quantifying the impact of occupant behaviors are both active areas of research. Here, we propose a data analytics framework for detecting changes in occupant behaviors, which will help build an analytics feedback loop from behavior impact to incentive design. The framework has two major parts. The first forecasts energy consumption for each occupant, while the second determines a probability distribution for changes in energy consumption. The parts are interchangeable with other existing machine learning and statistical methods. A specific instantiation of the framework, using kernel ridge-regression for forecasting and k-means to find an empirical behavior distribution, is described in detail. An HVAC use-case with 5 different incentivized behaviors is used as an example to show that the framework can detect behavior changes induced by incentives. Furthermore, we show that some simpler behavior-change detection methods do not work, further justifying the use of advanced analytics.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Extreme metrics from large ensembles: investigating the effects of ensemble size on their estimates

Abstract. We consider the problem of estimating the ensemble sizes required to characterize the forced component and the internal variability of a number of extreme metrics. While we exploit existing large ensembles, our perspective is that of a modeling center wanting to estimate a priori such sizes on the basis of an existing small ensemble (we assume the availability of only five members here). We therefore ask if such a small-size ensemble is sufficient to estimate accurately the population variance (i.e., the ensemble internal variability) and then apply a well-established formula that quantifies the expected error in the estimation of the population mean (i.e., the forced component) as a function of the sample size n, here taken to mean the ensemble size. We find that indeed we can anticipate errors in the estimation of the forced component for temperature and precipitation extremes as a function of n by plugging into the formula an estimate of the population variance derived on the basis of five members. For a range of spatial and temporal scales, forcing levels (we use simulations under Representative Concentration Pathway 8.5) and two models considered here as our proof of concept, it appears that an ensemble size of 20 or 25 members can provide estimates of the forced component for the extreme metrics considered that remain within small absolute and percentage errors. Additional members beyond 20 or 25 add only marginal precision to the estimate, and this remains true when statistical inference through extreme value analysis is used. We then ask about the ensemble size required to estimate the ensemble variance (a measure of internal variability) along the length of the simulation and – importantly – about the ensemble size required to detect significant changes in such variance along the simulation with increased external forcings. Using the F test, we find that estimates on the basis of only 5 or 10 ensemble members accurately represent the full ensemble variance even when the analysis is conducted at the grid-point scale. The detection of changes in the variance when comparing different times along the simulation, especially for the precipitation-based metrics, requires larger sizes but not larger than 15 or 20 members. While we recognize that there will always exist applications and metric definitions requiring larger statistical power and therefore ensemble sizes, our results suggest that for a wide range of analysis targets and scales an effective estimate of both forced component and internal variability can be achieved with sizes below 30 members. This invites consideration of the possibility of exploring additional sources of uncertainty, such as physics parameter settings, when designing ensemble simulations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Uncertainty-Aware, Structure-Preserving Machine Learning Approach for Domain Shift Detection From Nonlinear Dynamic Responses of Structural Systems

Complex structural systems deployed for aerospace, civil, or mechanical applications must operate reliably under varying operational conditions. Structural health monitoring (SHM) systems help ensure the reliability of these systems by providing continuous monitoring of the state of the structure. SHM relies on synthesizing measured data with a predictive model to make informed decisions about structural states. However, these models—which may be thought of as a form of a digital twin—need to be updated continuously as structural changes (e.g., due to damage) arise. We propose an uncertainty-aware machine learning model that enforces distance preservation of the original input state space and then encodes a distance-aware mechanism via a Gaussian process (GP) kernel. The proposed approach leverages the spectral-normalized neural GP algorithm to combine the flexibility of neural networks with the advantages of GP, subjected to structure-preserving constraints, to produce an uncertainty-aware model. This model is used to detect domain shift due to structural changes that cannot be observed directly because they may be spatially isolated (e.g., inside a joint or localized damage). This work leverages detection theory to detect domain shift systematically given statistical features of the prediction variance produced by the model. The proposed approach is demonstrated on a nonlinear structure being subjected to damage conditions. In conclusion, it is shown that the proposed approach is able to rely on distances of the transformed input state space to predict increased variance in shifted domains while being robust to normative changes.

Algorithms↗
DOE OSTI2024

Detection of astrophysical neutrinos at prospective locations of dark matter detectors

We study the prospects for detection of solar, atmospheric neutrino, and diffuse supernova neutrino background (DSNB) fluxes at future large-scale dark matter detectors through both electron and nuclear recoils. We specifically examine how the detection prospects change for several prospective detector locations [Sanford Underground Research Facility (SURF), SNOlab, Gran Sasso, China Jinping Underground Laboratory (CJPL), and Kamioka] and improve upon the statistical methodologies used in previous studies. Because of its ability to measure lower neutrino energies than other locations, we find that the best prospects for the atmospheric neutrino flux are at the SURF location, while the prospects are weakest at CJPL because it is restricted to higher neutrino energies. On the contrary, the prospects for the DSNB are best at CJPL, due largely to the reduced atmospheric neutrino background at this location. Including full detector resolution and efficiency models, the CNO component of the solar flux is detectable via the electron recoil channel with exposures of ∼ 10 3 ton-yr for all locations. These results highlight the benefits for employing two detector locations, one at high and one at low latitude. Published by the American Physical Society 2024

Astronomy & Astrophysics↗
DOE OSTI2021

Constraining and Characterizing the Size of Atmospheric Rivers: A Perspective Independent From the Detection Algorithm

Abstract Atmospheric rivers (AR) are large and narrow filaments of poleward horizontal water vapor transport. Because of its direct relationship with horizontal vapor transport, extreme precipitation, and overall AR impacts over land, the AR size is an important characteristic that needs to be better understood. Current AR detection and tracking algorithms have resulted in large uncertainty in estimating AR sizes, with areas varying over several orders of magnitude among different detection methods. We develop and implement five independent size estimation methods to characterize the size of ARs that make landfall over the west coast of North America in the 1980–2017 period and reduce the range of size estimation from ARTMIP. ARs that originate in the Northwest Pacific (WP) (100°−180°E) have larger sizes and are more zonally oriented than those from the Northeast Pacific (EP) (180°−240°E). ARs become smaller through their life cycle, mainly due to reductions in their width. They also become more meridionally oriented toward the end of their life cycle. Overall, the size estimation methods proposed in this study provide a range of AR areas (between 7 × 10 11 and 10 13 m 2 ), that is, several orders of magnitude narrower than current methods estimation. This methodology can provide statistical constraints in size and geometry for the AR detection and tracking algorithms, and an objective insight for future studies about AR size changes under different climate scenarios.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Experimental determination of precision, resolution, accuracy and trueness of time-of-flight neutron diffraction strain measurements

We discuss a simple statistical analysis which yields the precision, resolution, accuracy and trueness of diffraction-based lattice strain measurements. The procedure consists of measuring the thermal expansion induced in each component of an ideal, non-reacting, two-component, crystalline powder sample in situ. One component, with a high coefficient of thermal expansion (CTE), serves as an internal thermometer. The quantities of interest are obtained by determining the smallest, statistically significant, thermal lattice strain which can be detected through diffraction analysis in the second, low CTE, component in response to controlled temperature changes. This procedure also provides a robust check of the alignment of the diffraction system and is able to reveal the presence of systematic errors. The application of this technique to a time-of-flight engineering diffractometer/strain scanner is presented.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Target Detection via Cognitive Radars Using Change-Point Detection, Learning, and Adaptation

Many radar detection algorithms that assume a stationary environment (clutter) have been proposed and analyzed over the years. However, in practice, changes in the nonstationary environment can perturb the parameters of the clutter distribution, or even alter the clutter distribution family, which can greatly deteriorate the target detection capability. To avoid such potential performance degradation, cognitive radar systems are envisioned which are required to rapidly realize the nonstationarity, accurately learn the new characteristics of the environments, and adaptively update the detector. In this paper, aiming to develop a fully cognitive radar for target detection in nonstationary environments, we propose a unifying framework that integrates (i) change-point detection of clutter distributions by using a data-driven cumulative sum (CUSUM) algorithm and its extended version, (ii) learning/identification of clutter distribution by applying sparse theory and kernel density estimation methods, and (iii) adaptive target detection by automatically modifying the likelihood-ratio test and corresponding detection threshold. Further, with extensive numerical examples, we demonstrate the achieved improvements in detection performance due to the proposed framework in comparison to a nonadaptive case, an adaptive matched filter (AMF) method, and the clairvoyant case. Herein, we also use Wilcoxon rank-sum tests to evaluate the statistical significance of the performance improvements

42 ENGINEERING↗
DOE OSTI2023

In Situ Investigation of Upper Airway Occlusion in Sleep Disordered Breathing Using Ultrasonic Transducer Arrays

This work presents a novel application of ultrasound for the real-time, non-invasive investigation of occlusion of the upper airway during events of obstructive sleep apnea/hypopnea syndrome. It is hypothesized that ultrasonic pulses applied to the neck during apneic events produce spectral and temporal features that can detect apnea occurrence. Theoretical models of ultrasound propagation and an in vitro test were conducted to test this hypothesis in both transmission and reflection modes. Complete specifications and technical details of the system design and fabrication, which is mounted on each subject’s neck, are presented, including the methodology. Nine patients (seven male and two female, mean age of 42 years, with a range of 25 to 56 years, and body mass index 37.6 ± 6.6 kg/m2) were recruited for a full night study, which included simultaneous nocturnal polysomnography for the validation of the results. Nine temporal features and four spectral features were extracted from the envelope of the received pulse waveform. These were used to compute 26 metrics to quantify the changes in the ultrasonic waveforms between normal breathing and apneic events. The statistical analysis of the collected ultrasonic data showed that at least two or more of the proposed features could detect apneic events in all subjects. The findings establish the feasibility of the proposed method as a cost-effective and non-invasive OSAHS screening tool.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2022

Detecting hidden transient events in noisy nonlinear time-series

The information impulse function (IIF), running Variance, and local Hölder Exponent are three conceptually different time-series evaluation techniques. These techniques examine time-series for local changes in information content, statistical variation, and point-wise smoothness, respectively. Using simulated data emulating a randomly excited nonlinear dynamical system, this study interrogates the utility of each method to correctly differentiate a transient event from the background while simultaneously locating it in time. Computational experiments are designed and conducted to evaluate the efficacy of each technique by varying pulse size, time location, and noise level in time-series. Our findings reveal that, in most cases, the first instance of a transient event is more easily observed with the information-based approach of IIF than with the Variance and local Hölder Exponent methods. While our study highlights the unique strengths of each technique, the results suggest that very robust and reliable event detection for nonlinear systems producing noisy time-series data can be obtained by incorporating the IIF into the analysis.

42 ENGINEERING↗
DOE OSTI2026

Robust Measurement of Stellar Streams around the Milky Way: Correcting Spatially Variable Observational Selection Effects in Optical Imaging Surveys

Observations of density variations in stellar streams are a promising probe of low-mass dark matter substructure in the Milky Way. However, survey systematics such as variations in seeing and sky brightness can also induce artificial fluctuations in the observed densities of known stellar streams. These variations arise because survey conditions affect both object detection and star–galaxy misclassification rates. To mitigate these effects, we use Balrog synthetic source injections in the Dark Energy Survey (DES) Y3 data to calculate detection rate variations and classification rates as functions of survey properties. We show that these rates are nearly separable with respect to survey properties and can be estimated with sufficient statistics from the synthetic catalogs. Applying these corrections reduces the standard deviation of relative detection rates across the DES footprint by a factor of 5, and our corrections significantly change the inferred linear density of the Phoenix stream when including faint objects. Additionally, for artificial streams with DES-like survey properties we are able to recover density power spectra with reduced bias. We also find that uncorrected power-spectrum results for Legacy Survey of Space and Time (LSST)-like data can be around 5 times more biased, highlighting the need for such corrections in future ground-based surveys.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2020

A Rapid Population Assessment Method for Wild Pigs Using Baited Cameras at 3 Study Sites

ABSTRACT Reliable and efficient population estimates are a critical need for effective management of invasive wild pigs ( Sus scrofa ). We evaluated the use of 10‐day camera grids for rapid population assessment (RPA) of wild pigs at 3 study sites that varied in vegetation communities and wild pig densities. Study areas included Buck Island Ranch, Florida; Tejon Ranch, California; and the Savannah River Site, South Carolina, USA, during 2016–2018. Rapid population assessments grids were composed of baited camera traps spaced approximately 500 or 750 m apart. Two RPA grids were deployed per study site and each grid was deployed twice (4–6 months apart) to assess changes in response to season or population control efforts. We assessed the ability of RPA grids to track population trends, how camera number influenced estimate precision, and how relative abundance indices related to density estimates. We detected changes in occupancy probability, detection probability, and N‐mixture estimates following removal operations and between seasons, but the ability of RPA grids to track population trends was dependent on the statistical method used and number of cameras traps. Increasing the number of cameras traps used in RPA grids increased precision, and these results can be used in determining survey design and estimate choice. We found that estimates of occupancy probability, detection probability, and N‐mixture estimates were positively correlated with spatially explicit capture–recapture density estimates. Thus, these less labor‐intensive estimates from RPA grids showed potential to index the relative abundance of wild pigs in some systems. Our evaluation of RPAs indicates that using study‐specific combinations of statistical method and number of cameras can provide a useful tool for monitoring wild pig presence, tracking population trends, and evaluating the effectiveness of management actions. © 2020 The Wildlife Society.

Schlichting, Peter E.↗
DOE OSTI2023

Impact of Glass Irradiation on Laser-Induced Breakdown Spectroscopy Data Analysis

Increased absorption of optical materials arising from exposure to ionizing radiation must be accounted for to accurately analyze laser-induced breakdown spectroscopy (LIBS) data retrieved from high-radiation environments. We evaluate this effect on two examples that mimic the diagnostics placed within novel nuclear reactor designs. The analysis is performed on LIBS data measured with 1% Xe gas in an ambient He environment and 1% Eu in a molten LiCl-KCl matrix, along with the measured optical absorption from the gamma- and neutron-irradiated low-OH fused silica and sapphire glasses. Significant changes in the number of laser shots required to reach a 3σ detection level are observed for the Eu data, increasing by two orders of magnitude after exposure to a 1.7 × 1017 n/cm2 neutron fluence. For all cases examined, the spectral dependence of absorption results in the introduction of systematic errors. Moreover, if lines from different spectral regions are used to create Boltzmann plots, this attenuation leads to statistically significant changes in the temperatures calculated from the Xe II lines and Eu II lines, lowering them from 8000 ± 610 K to 6900 ± 810 K and from 15,800 ± 400 K to 7200 ± 800 K, respectively, for exposure to the 1.7 × 1017 n/cm2 fluence. The temperature range required for a 95% confidence interval for the calculated temperature is also broadened. In the case of measuring the Xe spectrum, these effects may be mitigated using only the longer-wavelength spectral region, where radiation attenuation is relatively small, or through analysis using the iterative Saha–Boltzmann method.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

Granger causal inference for climate change attribution

Abstract Climate change detection and attribution (D&A) is concerned with determining the extent to which anthropogenic activities have influenced specific aspects of the global climate system. D&A fits within the broader field of causal inference, the collection of statistical methods that identify cause and effect relationships. There are a wide variety of methods for making attribution statements, each of which require different types of input data and focus on different types of weather and climate events and each of which are conditional to varying extents. Some methods are based on Pearl causality (direct experimental interference) while others leverage Granger (predictive) causality, and the causal framing provides important context for how the resulting attribution conclusion should be interpreted. However, while Granger-causal attribution analyses have become more common, there is no clear statement of their strengths and weaknesses relative to Pearl-causal attribution and no clear consensus on where and when Granger-causal perspectives are appropriate. In this prospective paper, we provide a formal definition for Granger-based approaches to trend and event attribution and a clear comparison with more traditional methods for assessing the human influence on extreme weather and climate events. Broadly speaking, Granger-causal attribution statements can be constructed quickly from observations and do not require computationally-intesive dynamical experiments. These analyses also enable rapid attribution, which is useful in the aftermath of a severe weather event, and provide multiple lines of evidence for anthropogenic climate change when paired with Pearl-causal attribution. Confidence in attribution statements is increased when different methodologies arrive at similar conclusions. Moving forward, we encourage the D&A community to embrace hybrid approaches to climate change attribution that leverage the strengths of both Granger and Pearl causality.

Risser, Mark D. (ORCID:0000000319561783)↗
DOE OSTI2022

Impact of Glass Irradiation on Laser-induced Breakdown Spectroscopy Diagnostics in the Visible and NIR Range

Increased absorption of optical materials arising from the exposure to ionizing radiation must be accounted for to accurately analyze the laser-induced breakdown spectroscopy (LIBS) data retrieved from high-radiation environments. We evaluate this effect on a specifc example that mimics the diagnostics placed within a gas-cooled fast reactor coolant stream. Analysis is performed on the LIBS data measured with 1% Xe gas in an ambient He environment and the measured optical absorption from the gamma- and neutron-irradiated low-OH fused silica and sapphire glasses. No significant change in the number of shots required to reach a 3σ detection level was observed up to 10 Mrad gamma dose and exposure to a 1.7×10 17 n/cm 2 neutron fluence. In contrast, the spectral dependence of absorption results in either a systematic overestimation or underestimation of line intensity ratios, depending on the line of interest selected for analysis. Moreover, if lines from different spectral regions are used to create Boltzmann plots, this attenuation also leads to statistically significant changes in the calculated temperatures for calculations using Xe II lines only, lowering them from 8000±610 K to 6800±810 K for the case of exposure to the 1.7×10 17 n/cm 2 flux, and broadening the temperature range required for a 95% confidence interval. In the case of measuring the Xe spectrum, these effects may be mitigated using only the longer-wavelength spectral region, where radiation-attenuation is relatively small, or through analysis using the iterative Saha-Boltzmann method.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

The dark energy survey: detection of weak lensing magnification of supernovae and constraints on dark matter haloes

The residuals of the distance moduli of Type Ia supernovae (SNe Ia) relative to a Hubble diagram fit contain information about the inhomogeneity of the Universe, due to weak lensing magnification by foreground matter. By correlating the residuals of the Dark Energy Survey Year 5 SN Ia sample (DES-SN5YR) with extragalactic foregrounds from the DES Y3 Gold catalogue, we detect the presence of lensing at |$6.0 \sigma$| significance. This is the first detection with a significance level above |$5\sigma$|⁠. Constraints on the effective mass-to-light ratios and radial profiles of dark matter haloes surrounding individual galaxies are also obtained. We show that the scatter of SNe Ia around the Hubble diagram is reduced by modifying the standardization of the distance moduli to include an easily calculable de-lensing (i.e. environmental) term. We use the de-lensed distance moduli to recompute cosmological parameters derived from SN Ia, finding in Flat wcold dark matter a difference of |$\Delta \Omega _{\rm M} = +0.036$| and |$\Delta w = -0.056$| compared to the unmodified distance moduli, a change of |$\sim 0.3\sigma$|⁠. We argue that our modelling of SN Ia lensing will lower systematics on future surveys with higher statistical power. We use the observed dispersion of lensing in DES-SN5YR to constrain |$\sigma _8$|⁠, but caution that the fit is sensitive to uncertainties at small scales. Nevertheless, our detection of SN Ia lensing opens a new pathway to study matter inhomogeneity that complements galaxy–galaxy lensing surveys and has unrelated systematics.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

Probing and Tuning Strain‐Localized Exciton Emission in 2D Material Bubbles at Room Temperature

In monolayer transition metal dichalcogenides bubbles-nanoscale deformations typically exhibiting a dome-like shape-Excitons are confined by the strain effect, which exhibits extraordinary emission properties, such as single photon generation, enhanced light emission, and spectrally tunable excitonic states. While the strain profiles of these bubbles are extensively studied, this work provides an approach 1) to directly visualize the associated exciton properties in bubbles formed in WSe2 monolayer, revealing an intrinsic emission wavelength shift of ≈40 nm, and 2) actively modify local strain, enabling further exciton emission tuning over a range of 50 nm. These are achieved by emission mapping and nanoindentation using a dielectric near-field probe, which enables the detection of local emission spectra and emission lifetimes within individual bubbles. Statistical analysis of 67 bubbles uncovers an emission wavelength distribution centered around 780 nm. Furthermore, saturation behavior in the power-dependent studies and the associated lifetime change reveal the localized nature of the strain-induced states. These findings provide direct insights into the strain-localized emission dynamics in bubbles and establish a robust framework for non-destructive, reversible, and predictable nanoscale emission control, presenting a potential avenue for developing next-generation tunable quantum optical sources.

2D materials bubbles↗
DOE OSTI2022

ThunderSecure: deploying real-time intrusion detection for 100G research networks by leveraging stream-based features and one-class classification network

Nowadays, data generated by large-scale scientific experiments are on the scale of petabytes per month. These data are transferred through dedicated high-bandwidth networks (40/100G) across distributed sites for processing, storage, and analysis. Like general purpose networks, research networks experience intrusions. However, monitoring anomalies in such high-speed network traffics is challenging given current cyber-infrastructure. Moreover, traditional network intrusion detection systems (NIDS) are signature based. However, anomaly patterns are difficult to define and that rulesets are often not updated frequently enough to reflect the changes of attack behaviors. We present ThunderSecure, a high-throughput, unsupervised learning-based intrusions detection system for 100G research networks. ThunderSecure implements an efficient packet processing and detection pipeline using multi-cores and GPUs. It extracts statistical and temporal features from real-time network data streams and feeds them to a one-class anomaly detection network. A baseline of normal distribution will be created based on the training observation. Testing traffic deviated from the learned profile will be marked as anomalies. We trained ThunderSecure on hundreds of billions of science data packets mirrored from two 100G network connections at Fermi National Accelerator Laboratory. The detection performance was evaluated on traffic captured from the same research network days and weeks after the training with different types of attack flows injected. Results show that ThunderSecure can recognize science data traffic captured long after the training and made nearly certain detection on the segment of the streams where anomalous flows were injected.

100G research network↗